?? http:^^fas.sfu.ca^cs^research^groups^vision^
字號:
Date: Tue, 26 Nov 1996 00:39:21 GMTServer: NCSA/1.5Content-type: text/htmlLast-modified: Mon, 27 May 1996 17:20:17 GMTContent-length: 7270<HTML><HEAD><TITLE> The Computational Vision Lab</TITLE><LINK REV="MADE" HREF="mailto:webmaster@cs.sfu.ca"></HEAD><BODY><CENTER><H1>The Computational Vision Lab</H1><HR><CENTER><!WA0><IMG SRC="http://fas.sfu.ca/cs/research/groups/Vision/Images/eyes402-158.jpg" WIDTH="402" HEIGHT="158" ALT="eyes.logo"></CENTER><HR><P><tr><td align="right"> School of Computing Science, <td width="10"></td></td><td align="left"> Fax: (604) 291-3045, </td></tr><tr><td align="right">Simon Fraser University <td width="10"></td></td><td align="left"> email: funt@cs.sfu.ca </td></tr><tr><td align="right"> Burnaby, B.C. V5A 1S6 Canada <td width="10"></td></td><td align="left"> Physical location: ASB 10865, SFU </td></tr><tr ><td><br></td></tr><tr><td align="right"> <!WA1><IMG SRC="http://fas.sfu.ca/css/graphics/small-css-logo.gif" WIDTH="54" HEIGHT="52"> </td><td width="10"></td><td align="left"> A <!WA2><A HREF="http://fas.sfu.ca/css">Centre for Systems Science</A> <br> Affiliated Laboratory </td></tr><br></table></CENTER><HR>Members of the Computational Vision Lab at SFU conduct research into machine vision and image processing, with emphasis on computational modelsof colour vision.<P>Dr. <!WA3><A HREF="http://fas.sfu.ca/cs/people/Faculty/Funt/index.html">Brian Funt</A>is the director of the lab.<HR><H2><!WA4><IMG ALIGN=LEFT SRC="http://fas.sfu.ca/css/graphics/rightmarblearrow.gif" WIDTH="30" HEIGHT="30">Overview </H2><P><CENTER><TABLE ALIGN="CENTER" COLSPACING="10" Border="0"><TR><TD><!WA5><IMG ALIGN=LEFT,MIDDLE SRC="http://fas.sfu.ca/cs/research/groups/Vision/Images/a2.0.C.gif" WIDTH="84" HEIGHT="132"> </TD><TD><!WA6><IMG ALIGN=RIGHT,MIDDLE SRC="http://fas.sfu.ca/cs/research/groups/Vision/Images/a2_track.0.gif" WIDTH="84" HEIGHT="132"> </TD></TR></TABLE><br><!WA7><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/gesture.html"><H5>Images for Automated Gesture Tracking </A></H5></CENTER><P> <EM>Computational Vision </EM>can be thought of as enablingcomputers to use visual information. Like many problems in ArtificialIntelligence, it's something people do so easily they barely thinkabout it, but a very complex problem for a machine. <P>Our primary focus in the Vision Lab at SFU is in understandingcolour: How are colours perceived? How can colours be reproduced accuratelyon different media? In what ways does colour help in understanding images? Understanding colour is a much more difficult problem than most peoplesuspect. Often poor colour rendition results more from our limitedunderstanding of colour perception than it does from limitations of ourcolour producing devices. <p>We subscribe to a computational view of colour; namely, that humanperception and use of colour can be explored and explained ascomputations. The fundamental problem of colour is to explain how wesee colours as relatively stable despite the fact that the lightreflected into our eyes from an object varies dramatically with thelight illuminating the object. Colour and computers have become muchmore intertwined in recent years as colour displays and colourprinters have become more affordable. Since colour is a perceptual,not a physical quality, it is crucial to have a good model of how weperceive colour in complex environments if we are to get predictableresults from these devices.<p>Colours are difficult to reproduce correctly, but why? While we'veall experienced untrue colour while using home video cameras orviewing prints from our local photofinisher, now we have colourprinters frustrating us with colours that look very little like thenice colours we previewed on our CRTs. When the colour doesn't lookright, it's natural to feel that the printer and monitor are notcalibrated properly--- and of course perhaps they're not--- but that'snot the fundamental problem. The fundamental problem stems from thefact that colour reproduction, simply is not a matter of reproducingidentical physical phenomena, as it is in the case of soundreproduction in which a similar pattern of sound waves is recreated,but rather a matter of creating perceptual equivalences.<p>For us to build machines that reproduce colours accurately or to makeeffective use of colour in robotics requires that we understand human colourperception; and the last decade has produced many interesting newcomputational theories of colour coming from both computer science andpsychology. A central concern of these theories is to describe how colourdepends or does not depend on the incident illumination. A coloured surfacecannot be seen unless we shine some light on it, but the spectrum of thereflected light depends on the product of the spectrum of the incidentlight's spectrum and the surface's reflectance. It's natural to think of asurface's colour as a feature of the surface itself, but the spectrum of thelight energy reaching the eye has the two factors of illumination andreflectance confounded into one. In order to determine the true surfaceproperties, the effect of the illumination must be taken into account. <P> <h2> Colour Correction Results </H2><P><CENTER><!WA8><IMG SRC="http://fas.sfu.ca/cs/research/groups/Vision/Images/result-back-red-small.jpg" WIDTH="303" HEIGHT="226" ALT="correction.results"></CENTER><P> In the upper left is the image of a scene taken under an orangish,tungsten light which has the effect of turning the background overlygreen and the whites a bit yellow. In the bottom right is the targetimage of the same scene but under the standard illumination for whichthe camera is calibrated. Since we cannot always control theillumination, our goal is to correct automatically the colours in theinput image so that they will look like those in the targetimage. (The fuzziness of the images is due to a high JPEG compressionfactor--- concentrate on the colours.)<P> The top right is a 'corrected' image produced by the standardgrey-world colour balancing algorithm, which assumes that the averageof all the colours in the scene is grey. In the bottom left is themuch better result produced by our new, more sophisticated algorithmdeveloped in the Vision Lab. As you can see, the result of our newalgorithm is much closer to the target image than that produced bythe grey-world algorithm.<P><!WA9><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/cgi/imagemap.cgi/bbar2"><!WA10><IMG SRC="http://fas.sfu.ca/cs/research/groups/Vision/Images/bbar2.gif" BORDER=0 ISMAP WIDTH="468" HEIGHT="68" ALT="Menubar" ></A><HR><TABLE border= 0><CAPTION> <H2> Other Links </H2> </CAPTION><TR> <TD></TD> <TD WIDTH="10"></TD> <TD> <!WA11><IMG SRC="http://fas.sfu.ca/cs/research/groups/Vision/Images/sfucrest.gif" width="81" height="110"> </TD> <TD WIDTH="10"></TD> <TD> <!WA12><IMG ALIGN=MIDDLE SRC=http://fas.sfu.ca/images/CMPTlogo-halfsize.gif WIDTH="141" HEIGHT="77"> </TD> </TR> <TR> <TD> Dynamic <br> Site <br> <!WA13><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/indexform.cgi">Index</A> </TD> <TD WIDTH="10"></TD> <TD> <!WA14><A HREF="http://www.sfu.ca/"> Simon Fraser University <br> Home Page</A></TD> <TD WIDTH="10"></TD> <TD> <!WA15><A HREF="http://fas.sfu.ca/cs"> SFU Computing Science <br> Home Page</A> </TD> </TR></TABLE><HR><P><FONT SIZE="-1">Text-only links:<!WA16><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/members.html"> Lab Members </A> | <!WA17><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/Pubs/publications.html"> Publications On-line </A> | <!WA18><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/Vislab-postin/mail2.html"> Mail Feedback </A> | <!WA19><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/links.html"> Links of Interest </A></FONT><P><H5>Page maintained by: <!WA20><A HREF="http://fas.sfu.ca/cs/research/groups/Vision/Brock/brock.html">Michael Brockington </A>- <!WA21><A HREF="mailto:brocking@sfu.ca">brocking@sfu.ca</A> - May 1996 </H5></BODY></HTML>?? 快捷鍵說明
復(fù)制代碼
Ctrl + C
搜索代碼
Ctrl + F
全屏模式
F11
切換主題
Ctrl + Shift + D
顯示快捷鍵
?
增大字號
Ctrl + =
減小字號
Ctrl + -